Llama 3.3

Latest

Meta

•

Open Source

•

70B

# 31

Released

Dec 6, 2024

# 9

Knowledge Cutoff

Dec 23

Benchmarks

# 50

Code LiveBench

36.6%

# 35

Code LMArena

1232

# 38

GPQA Diamond

50.5%

# 29

Reason LiveBench

50.7%

# 37

ELO LMArena

1257

# 36

Math LiveBench

41.1%

# 20

MATH

77.0%

# 13

Human Eval

88.4%

# 2

Human Eval+

87.6%

# 22

NYT Connections

15.1%

# 19

Aider Old

59.4%

# 28

MMLU Pro

68.9%

# 14

MMLU

86.0%

# 17

Halluc. Hughes

4.0%

# 18

Aidan Bench

779

# 5

IF LiveBench

82.7%

# 37

Data LiveBench

48.3%

# 33

Lang LiveBench

39.2%

# 26

Avg LiveBench

56.7%

# 2

IF Evaluation

92.1%

Pricing

# 9

Input Cost /M

$0.1

# 12

Output Cost /M

$0.4